OpenSSH 7.0 contained a logic error in PermitRootLogin= prohibit-password/without-password that could, depending on compile-time configuration, permit password authentication to root while preventing other forms of authentication. This problem was reported by Mantas Mikulenas.

eq 7.0.p1,1

OpenSSH 6.8 and 6.9 incorrectly set TTYs to be world-writable. Local attackers may be able to write arbitrary messages to logged-in users, including terminal escape sequences. Reported by Nikolay Edigaryev.

Fixed a privilege separation weakness related to PAM support. Attackers who could successfully compromise the pre-authentication process for remote code execution and who had valid credentials on the host could impersonate other users.

Fixed a use-after-free bug related to PAM support that was reachable by attackers who could compromise the pre-authentication process for remote code execution.

< 7.0.p1,1

ge 10.2 lt 10.2_2

ge 10.1 lt 10.1_19

ge 9.3 lt 9.3_24

OpenBSD Project reports:

sshd(8) from OpenSSH 6.2 through 8.7 failed to correctly initialise supplemental groups when executing an AuthorizedKeysCommand or AuthorizedPrincipalsCommand, where a AuthorizedKeysCommandUser or AuthorizedPrincipalsCommandUser directive has been set to run the command as a different user. Instead these commands would inherit the groups that sshd(8) was started with.

Depending on system configuration, inherited groups may allow AuthorizedKeysCommand/AuthorizedPrincipalsCommand helper programs to gain unintended privilege.

Neither AuthorizedKeysCommand nor AuthorizedPrincipalsCommand are enabled by default in sshd_config(5).

ge 6.2.p1,1 lt 8.7.p1_2,1

Problem Description:

The ssh-agent(1) agent supports loading a PKCS#11 module from outside a trusted whitelist. An attacker can request loading of a PKCS#11 module across forwarded agent-socket. [CVE-2016-10009]

When privilege separation is disabled, forwarded Unix domain sockets would be created by sshd(8) with the privileges of 'root' instead of the authenticated user. [CVE-2016-10010]

Impact:

A remote attacker who have control of a forwarded agent-socket on a remote system and have the ability to write files on the system running ssh-agent(1) agent can run arbitrary code under the same user credential. Because the attacker must already have some control on both systems, it is relatively hard to exploit this vulnerability in a practical attack. [CVE-2016-10009]

When privilege separation is disabled (on FreeBSD, privilege separation is enabled by default and has to be explicitly disabled), an authenticated attacker can potentially gain root privileges on systems running OpenSSH server. [CVE-2016-10010]

< 7.3.p1_5,1

ge 11.0 lt 11.0_7

ge 10.3 lt 10.3_16

Problem Description

The CRC compensation attack detector in the sshd(8) daemon, upon receipt of duplicate blocks, uses CPU time cubic in the number of duplicate blocks received. [CVE-2006-4924]

A race condition exists in a signal handler used by the sshd(8) daemon to handle the LoginGraceTime option, which can potentially cause some cleanup routines to be executed multiple times. [CVE-2006-5051]

Impact

An attacker sending specially crafted packets to sshd(8) can cause a Denial of Service by using 100% of CPU time until a connection timeout occurs. Since this attack can be performed over multiple connections simultaneously, it is possible to cause up to MaxStartups (10 by default) sshd processes to use all the CPU time they can obtain. [CVE-2006-4924]

The OpenSSH project believe that the race condition can lead to a Denial of Service or potentially remote code execution, but the FreeBSD Security Team has been unable to verify the exact impact. [CVE-2006-5051]

Workaround

The attack against the CRC compensation attack detector can be avoided by disabling SSH Protocol version 1 support in sshd_config(5).

There is no workaround for the second issue.

ge 6.1 lt 6.1_10

ge 6.0 lt 6.0_15

ge 5.5 lt 5.5_8

ge 5.4 lt 5.4_22

ge 5.0 lt 5.3_37

< 4.11_25

< 4.4,1

< 4.4.p1,1

The OpenSSH development team reports:

A memory corruption vulnerability exists in the post- authentication sshd process when an AES-GCM cipher (aes128-gcm@openssh.com or aes256-gcm@openssh.com) is selected during kex exchange.

If exploited, this vulnerability might permit code execution with the privileges of the authenticated user and may therefore allow bypassing restricted shell/command configurations.

Either upgrade to 6.4 or disable AES-GCM in the server configuration. The following sshd_config option will disable AES-GCM while leaving other ciphers active:

Ciphers aes128-ctr,aes192-ctr,aes256-ctr,aes128-cbc,3des-cbc,blowfish-cbc,cast128-cbc,aes192-cbc,aes256-cbc

ge 6.2.p2,1 lt 6.4.p1,1

ge 6.2.p2,1 lt 6.4.p1,1

It was discovered that the OpenSSH sshd daemon did not check the list of keyboard-interactive authentication methods for duplicates. A remote attacker could use this flaw to bypass the MaxAuthTries limit, making it easier to perform password guessing attacks.

< 6.9.p1_2,1

ge 10.1 lt 10.1_16

ge 9.3 lt 9.3_21

ge 8.4 lt 8.4_36

Problem Description:

When processing the SSH_MSG_KEXINIT message, the server could allocate up to a few hundreds of megabytes of memory per each connection, before any authentication take place.

Impact:

A remote attacker may be able to cause a SSH server to allocate an excessive amount of memory. Note that the default MaxStartups setting on FreeBSD will limit the effectiveness of this attack.

< 7.3p1_1

ge 11.0 lt 11.0_3

ge 10.3 lt 10.3_12

OpenBSD Project reports:

ssh-agent(1): fixed a double-free memory corruption that was introduced in OpenSSH 8.2 . We treat all such memory faults as potentially exploitable. This bug could be reached by an attacker with access to the agent socket.

On modern operating systems where the OS can provide information about the user identity connected to a socket, OpenSSH ssh-agent and sshd limit agent socket access only to the originating user and root. Additional mitigation may be afforded by the system's malloc(3)/free(3) implementation, if it detects double-free conditions.

The most likely scenario for exploitation is a user forwarding an agent either to an account shared with a malicious user or to a host with an attacker holding root access.

ge 8.2.p1,1 lt 8.4.p1_4,1

The OpenSSH project reports:

* sshd(8): Mitigate timing differences in password authentication that could be used to discern valid from invalid account names when long passwords were sent and particular password hashing algorithms are in use on the server. CVE-2016-6210, reported by EddieEzra.Harari at verint.com

* sshd(8): (portable only) Ignore PAM environment vars when UseLogin=yes. If PAM is configured to read user-specified environment variables and UseLogin=yes in sshd_config, then a hostile local user may attack /bin/login via LD_PRELOAD or similar environment variables set via PAM. CVE-2015-8325, found by Shayan Sadigh.

< 7.3.p1,1

OpenSSH reports:

OpenSSH clients between versions 5.4 and 7.1 are vulnerable to information disclosure that may allow a malicious server to retrieve information including under some circumstances, user's private keys.

gt 5.4.p0,1 lt 7.1.p2,1

ge 10.2 lt 10.2_10

ge 10.1 lt 10.1_27

ge 9.3 lt 9.3_34

The OpenSSH project reports:

Missing sanitisation of untrusted input allows an authenticated user who is able to request X11 forwarding to inject commands to xauth(1).

Injection of xauth commands grants the ability to read arbitrary files under the authenticated user's privilege, Other xauth commands allow limited information leakage, file overwrite, port probing and generally expose xauth(1), which was not written with a hostile user in mind, as an attack surface.

Mitigation:

Set X11Forwarding=no in sshd_config. This is the default.

For authorized_keys that specify a "command" restriction, also set the "restrict" (available in OpenSSH >=7.2) or "no-x11-forwarding" restrictions.

< 7.2.p2,1

ge 10.2 lt 10.2_14

ge 10.1 lt 10.1_31

ge 9.3 lt 9.3_39